Fully-Connected-Based Adaptive Speckles Optimization Method for Ghost Imaging

Ghost imaging (GI) has recently emerged as a promising new imaging technology, generating considerable interest in the field. Using appropriate speckles in GI can result in higher-quality reconstructed target images. The conventional method of projecting random speckles to illuminate unknown targets often results in low efficiency and poor image reconstruction quality. Recently, we find that speckles in GI act similarly to fully-connected layers in neural networks, in terms of information processing and transmission. Therefore, we propose an improved projection method for speckles, called AGOS, which is based on adaptive gradient optimization employing a fully-connected layer. Demonstrations based on simulations and experiments show that the proposed method in this letter can achieve better results than random speckles under the same conditions.